-
Competitor rules
Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.
AMD VEGA confirmed for 2017 H1
- Thread starter Doom112
- Start date
- Status
2 versions of Vega, easily the reason for the confusion.
The 1080 boosts very high automatically - I hope you're considering the boost clock to be the stock clock?
Either way in the vast majority of benchmarks I've seen, the 1080 is 40-50% faster than the Fury X.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 31,179
nVidia have probably known about Vega's performance for months now and are comfortable that Pascal refresh (as well as a Ti release) will be on top generally.
It's the cost that has me concerned for AMD tbh, it's obvious that HBM2 is not going to be cheap to implement! I'm sure they'll have some GDDR5X options for the lower tier card/s
Soldato
- Joined
- 19 Feb 2011
- Posts
- 5,849
While i somewhat agree with the cost of HBM2 affecting the price of the Vega top end cards and i dont expect them to be cheap, i just hope we get a minimum of 8GB of HBM2, we already know that 4GB Of HMB1 acts like roughly 6GB of GDDR5? from various comments i have seen, which theoretically means 8GB of HBM2 could potentially be around 12GB of GDDR5 with the compression tech etc?
What was holding Fiji back at lower res? was it the HBM or just the arch in general? as the cards got better the higher the res?
Anyhow, im in for either 2 small Vega or 1 big Vega all depending on price and performance, i expect big Vega to be within £70 or high end Nvidia (prob the 1080ti) if the performance is within 10%.
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,357
- Location
- Planet Earth
The latest TPU charts put it at 35% now.
I can assure you from personal experience that 4gb of HBM1 acts exactly the same as 4gb of GDDR5 capacity wise.
Also performance wise it is pretty bad compared to the GDDR5 cards @1080p, it is only @2160p that it really does shine.
Some marginal performance advantages when its shuffling a lot of data around but when its up against the wall 4GB is 4GB and the bottleneck is always going to be the system bus once you are swapping data in and out of VRAM due to space constraints - no way around that HBM or otherwise.
Soldato
- Joined
- 19 Feb 2011
- Posts
- 5,849
So is that a limitation of HBM or of the Fiji Architecture do you think? im starting to wonder what actual advantage HBM will have for a GPU? it cant really be OC'd, and it does not appear to be any better than GDDR? its a lot more expensive? what actual benefit does it have?
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,357
- Location
- Planet Earth
AMD said they were actively managing the VRAM on the Fury cards IIRC.
Caporegime
- Joined
- 8 Jul 2003
- Posts
- 30,063
- Location
- In a house
Yup, through their drivers.
Well, they said that they were hand tuning games for memory management but its hard to say how long that went on for.
Caporegime
- Joined
- 9 Nov 2009
- Posts
- 25,357
- Location
- Planet Earth
Probably still happening as the Fury cards seem to still do reasonably well for 4GB cards at higher resolutions.
Permabanned
- Joined
- 12 Sep 2013
- Posts
- 9,221
- Location
- Knowhere
When they told us that HBM would have to be managed on a game to game basis that was one of my concerns before buying my Fury, However I'm glad to see that they've done a pretty good job over the last year although my concern now is that once Vega is here will the 4gb HBM cards still get optimised for as well as they have been to date or will the support needed fizzle out as they get replaced?
I find it hard to believe that AMD will have the resources to support the Fury series VRAM optimisations for long.
Once Vega launches, I think Fury support will be completely neglected. The Fury cards were not popular, and current users can only be a very small percentage of current AMD users, so they won't upset that many people. Fiji will still have had ~2 years of full support.
Probably about as far as it does for GDDR5 and GDDR5X based cards.
This is something I would think that all drivers do from both vendors.
There are plenty of 4gb Hawaii cards out there, does their memory need hand tuning for individual games too?
Once their 8GB+ cards are out that supplant the Fury series as their top tier high resolution product, they will fizzle out. Realistically, in a year or two's time, Fury's are going to be mid tier equivalent cards at best and 4GB an okay match to their performance. I hope they don't drop it too quick because you don't want to burn buyers who went on a limb and purchased your last generation's (prototype) high end product.
An AMD interview alluded to 1-2 engineers (iirc) being capable of performing the memory management/optimisation, so it shouldn't be excessive to maintain. If those optimisations carry over to other gpu's then it might be justifiable (dunno if they do).
+1
Hey this is AMD NOT Nvidia
I actually don't think they will do that. Not for a good while anyway. Don't go tarring AMD with Nvidia's brush. AMD have always supported older cards far, far better than Nvidia ever have.
If they go down that route then you can get yer tar and yer brush out and I will say fair cop....but until then....@ £249 in the Black Friday sale it is the best bang for buck card out there.
I luvs my Fury, I does. It has served me well and is going great guns at the moment
Money has been saved though and a weary eye on Vega is being kept. Fingers crossed for AMD and a good competitive card in 2017.
as long cards are supported optimizations like this will go on.
Money has been saved though and a weary eye on Vega is being kept. Fingers crossed for AMD and a good competitive card in 2017.
Vega brightly awaits you.
best star and card in town soon.
Caporegime
- Joined
- 18 Oct 2002
- Posts
- 31,179
"A bird does not sing because it has an answer. It sings because it is insane"
- Status